nVidia 1080 ti

[biochemical_animations]

I read bad reviews about the risers...which ones did you go with?

[mitviz]

I read bad reviews about the risers...which ones did you go with?

i know people using them, i also use them now, there is nothing bad about it, the place where i bought my gpus even recommended it, here is the one i bought from this exact seller
http://www.ebay.com/itm/162273850492...%3AMEBIDX%3AIT

[biochemical_animations]

Thanks. Those are different then the ribbon ones I read about

[mitviz]

Thanks. Those are different then the ribbon ones I read about

these are usb 3.0 and alot better from what i heard

[biochemical_animations]

Yep, after thinking about it, I'm probably in for a new workstation with four of these in the near future. The scene setup, lighting, materials, etc workflow will be greatly speed up with a super fast gpu rendering WS. I just have to pitch a bit more for some upcoming jobs to cover the investment. Once done, tho, the will be nice. Keeping my fingers crossed.

This along with rt gpu reworking that Vlado mentioned will be, I believe, an incredible boost to photoreal workflow.

Exciting times.

[Raph4]

Another option I was considering, was putting a liquid cooler on the card that has the least airflow. I'm concerned tho for leaks, so I would probably design the thing where if there were leaks, they would not drip on anything electrical...so maybe rotating the case in some way so leaks would avoid components...just some ideas.

Right now, I'm still going with blowers...but liquid cooling would enable higher OCing.

I don't overclock my graphic cards, but with water block I have divided by 2 temperature of my 1080.
I think it's the same for 1080Ti.

[biochemical_animations]

i know people using them, i also use them now, there is nothing bad about it, the place where i bought my gpus even recommended it, here is the one i bought from this exact seller
http://www.ebay.com/itm/162273850492...%3AMEBIDX%3AIT

Are there differences in load times using the risers vs plugged straight in th MB?

I'm not totally up to knowledge on transfer speeds on the MB vs riser.

[mitviz]

Are there differences in load times using the risers vs plugged straight in th MB?

I'm not totally up to knowledge on transfer speeds on the MB vs riser.

no difference at all except that cards run much cooler

[iancamarillo]

hmmm. so if i bought 6 1070's and some power I could use all 6 cards? do you know if theres a limit?

[mitviz]

you should be able to use all 6, am at 4 now, waiting on the 1080 ti to make it 5

[iancamarillo]

nice. does that mean you'd be limited in ram usage to the older cards? Do you know if processor frequency affects GPU rendering? Or is it just more cuda cores and ram the better?

[mitviz]

nice. does that mean you'd be limited in ram usage to the older cards? Do you know if processor frequency affects GPU rendering? Or is it just more cuda cores and ram the better?

not sure what you mean by limited on older cards, older cards some of them cannot be used some just have too little vram for use anyway, faster processor means the scene is loaded faster so the gpu can take over but once it starts rendering the cpu doesnt matter much anymore the gpu takes over, more cuda cores and more vram cards are better, the 1080 ti is now the best buy. usually when am rendering with gpu i can work on other projects at the same time because they are more cpu driven and less gpu like for example i have a project going, i can let it render and then start another one at the same time

[iancamarillo]

Sorry I meant if you mix 1080's and 1080ti's, are you limited to the 1080's ram? And not utilizing all of the 1080ti's? I do a lot a arch renderings that take a LOT of ram and I want to use GPU in production but I'm not sure how much ram I need.

[Dariusz Makowski (Dadal)]

Sorry I meant if you mix 1080's and 1080ti's, are you limited to the 1080's ram? And not utilizing all of the 1080ti's? I do a lot a arch renderings that take a LOT of ram and I want to use GPU in production but I'm not sure how much ram I need.

I would suggest waiting for NVLink to be more available. Then u can share GPU ram...

[iancamarillo]

Yeh i read that article dabarti wrote - amazing. maybe the next gen gtx's will support nvlink. fingers crossed